Exercises: The Quantum Programming Career

Difficulty: ★ warm-up · ★★ standard · ★★★ extended · ★★★★ open-ended.

[audit] — apply the eight questions. [harvest] — derive the answer from the book's files rather than from memory. [career] — the answer is about you, and there is no key.

The last exercise set. Several of these ask you to check the book, and at least two of them will find something.

The scorecard (§40.3)

40.1 ★ [harvest] Run scripts/harvest_scorecard.py. How many lines in the book carry a measured number?

40.2 ★ Of the six head-to-head comparisons, how many did a quantum method win?

40.3 ★★ [harvest] Pick three numbers from §40.3 and find them in their source chapters. Report the section each came from.

40.4 ★★ Chapter 32's VQC tied LogisticRegression at 1.0000 on iris-binary. Explain why a tie there is weaker evidence than a loss elsewhere would be.

40.5 ★★ §40.3 lists "where things genuinely worked." Which entry is the strongest evidence for quantum computing, and which is the weakest? Defend both choices.

40.6 ★★★ [harvest] Extend harvest_scorecard.py to extract every number followed by a × or x multiplier. What is the largest claimed ratio in the book, and is it defensible?

40.7 ★★★ The scorecard splits results into four blocks. Propose a fifth category the book's findings support, and populate it from the chapters.

The checklist (§40.4)

40.8 ★ Why is "full CI" in TRIVIAL_BASELINES alongside "random"?

40.9 ★ Why does significant() raise on a standard error of exactly zero?

40.10 ★★ [audit] Write the Claim for a result you have produced — at work, in study, anywhere. Run audit_claim. Report the failures without softening them.

40.11 ★★ audit_claim returns failures rather than a score. Argue for a score, then against, then say which side Case Study 40.2 supports.

40.12 ★★ Question 7 is "can the measurement detect the thing being claimed?" Find the six Part V instances and say what each was blind to.

40.13 ★★★ [audit] Add a ninth check to vqelab.claims. Justify it from a chapter's finding, and say what it would have caught.

40.14 ★★★ The seven too-small measurements are listed in §40.4. Ch.39's is "different in kind." Explain how, and find a third kind of too-small that neither category covers.

40.15 ★★★ [audit] Take a quantum computing press release from the last year. Audit it. Then write the two-sentence version its authors could have defended.

Careers (§40.1–40.2, §40.6)

40.16 ★ Which of §40.1's roles has the largest headcount, and which the smallest?

40.17 ★★ §40.2 says deep circuit-identity manipulation "does not transfer." Argue against that, then say what would have to be true for you to be right.

40.18 ★★ [career] Which three of §40.2's transferable skills do you already have? Which one would you have to build first?

40.19 ★★ Case Study 40.1's engineer chose the offer with a bounded downside. Construct the case for Offer A that a different person could reasonably act on.

40.20 ★★★ [career] Write the five questions you would ask in a quantum computing interview to determine whether the team's benchmark could return a negative result. Rank them by how likely they are to get an honest answer.

40.21 ★★★ §40.6 says "a research programme with no failure condition is a marketing programme." Apply that to a project you have worked on. What was the failure condition, and was it real?

40.22 ★★★★ [open] [career] Find three current job postings with "quantum" in the title. For each, determine which of §40.1's roles it actually is. Report how many were described accurately.

Where the real problems are (§40.5)

40.23 ★★ §40.5 names three areas that survive the book's scrutiny. For each, name the chapter whose measurement supports it.

40.24 ★★ [harvest] Chapter 15 measured 450 → 2,882 physical qubits for one T gate. Find the figure for ten T gates, and say whether the cliff or the slope matters more.

40.25 ★★★ §40.5 says optimization, finance, and near-term ML are "furthest behind." Construct the strongest counterargument, using something in the book.

40.26 ★★★ Chapter 35 identified learning from quantum data as the surviving QML case. What would a convincing demonstration look like, and what would distinguish it from Chapter 35's own measurement?

40.27 ★★★★ [open] Pick one of §40.5's three areas and read its literature for a month. Report what you found that the book got wrong.

Closing

40.28 ★★ [harvest] Run project-checkpoint.py in this chapter. What does test_every_scorecard_number_is_in_its_source_chapter protect against, and what would break it?

40.29 ★★ Case Study 40.2 finds two of the book's four headline results failing its own checklist. Is that a reason to trust the book more or less?

40.30 ★★★ [audit] Audit a claim from Part I or II — the chapters written before the book's methodology had fully developed. Are they weaker?

40.31 ★★★ §40.7 says "none of those are quantum lessons." Name the field you work in and translate three of them into it.

40.32 ★★★ The book's refrain is that the easy number is the flattering one. Find a counterexample — a case where the cheap measurement is harsher than the careful one — and say what makes it different.

40.33 ★★★★ [open] Reproduce any measurement in this book on hardware you have access to. Report the difference and diagnose it. Chapter 39 §39.8 lists what to record.

40.34 ★★★★ [open] [career] Write the chapter this book should have had and did not. One page, with at least one measurement.

40.35 ★★★★ [open] In two years, re-read §40.3. Which entries have changed? That list is the field's actual rate of progress, and it is the only honest way to measure it.